I’m Marc Tombroff and This Is How I Mesh

Hi, I’m Marc Tombroff and I’m a VP Research & Development CFD at Cadence. I am an engineer and since my time at the engineering school I have always been impressed by the technology, its innovation drivers with the feeling that life would be fun if one could spend lifetime and work with researchers and creative people.

Experience is a combination of hard work and opportunities (other would say luck). I was lucky to meet Prof. Dr. Charles Hirsch, first as a student, and then as my employer when he offered me to become his first employee at NUMECA. I was 25 years old, holding a Master’s Degree in aerodynamics and CFD from the University of Brussels and a Master’s in business & administration from the Solvay Business School in Brussels, and some earlier short term employment experience in the world of large engineering and finance groups.

When Charles made this proposal in 1993, I did not hesitate long, resigned from my position at JPMorgan as software analyst, and jumped into the CFD business. At that time, they were only a few commercial CFD codes on the market such as Fluent and TascFlow, among some others.

Charles created NUMECA as a spin-off of the fluid mechanics lab of the Vrije Universiteit Brussels. As the company’s first employee, I had to split my work among different activities, from engineering services, sales and marketing, tech support, product definition, and so on. The first software we commercialized was IGG, a geometry modeler and grid generation system. IGG remains today the backbone of our flagship AutoGrid mesh generator. I remember very well these early times at NUMECA, meshing quite complex geometries for the European Propulsion Company, in charge of the design of the Vulcain engine of Ariane 5, spending weeks to get high quality mesh of the liquid hydrogen turbo pump and its radial impeller, before running the Euranus solver (our first code, that is still in use, though largely refactored since then, in our OMNIS™/Turbo suite).

Examples of block structured meshes created with the first version of IGG/AutoGrid years ago.

It is amazing to see that today, the same exercise takes a few minutes for the meshing and another few minutes for running the solver on parallel computer, in a fully batch and automatic workflow.

In 1998 I became General Manager of NUMECA. Since then, I do less meshing and CFD simulations and focus more on growth, bringing NUMECA to the next levels, with a growing team of extraordinary, talented researchers, engineers, and friends.

  • Location: Brussels, Belgium
  • Current position: VP Research & Development CFD at Cadence
  • Current computer: Macbook
  • One word that best describes how you work: Eagerly

When did you first become aware of Pointwise?

The first time I met Pointwise was in Japan in the 1990s where I met John Chawner and Rick Matus on several occasions of user meetings and workshops. The situation was odd, as despite we were competitors, it happened that we had the same distributor in Japan!

[John] I remember our meetings in Tokyo at the big VINAS Users Conference (the annual meeting of our Japanese distributor). Those trips are memorable because we had an excellent host. And I don’t think either of us could’ve predicted back then that we’d become colleagues.

What do you see are the biggest challenges facing CFD in the next 5 years?

CFD is extremely complex by nature, having to simulate multitude of different complex nonlinear physics, dealing with complex detailed CAD geometries and solving PDEs-based numerical models requiring heavy computer resources. Industry however requires easy-to-use tools, reliable and fast return time.

Significant progress has been made in response to these challenges over the years, and CFD methodologies are now widely accepted and applied with a strong impact in industry. The different CFD components have reached a high level of maturity, with high quality grid generation tools, robust CFD solvers with large range of physical models, post-processing and MDO tools for data analysis and optimization.

However, CFD predictions are still subject to many limitations, associated to the description of turbulence and transition, to the uncertainties related to geometry simplification, to the scaling on large hybrid HPC architecture, taking advantage of the most advanced CPU/GPU hardware while maintaining unique software architecture in a simple and integrated customer experience.

What are you currently working on?

We work on these quite exciting challenges, developing new solver technology to remove current limitation of turbulence and transition, hybrid GPU/CPU architecture, distributed parallel & client server grid generation and co-processing tools, and advanced software environment offering one unique software experience supporting multidisciplinary solvers and technologies in response to all type of CFD applications.

Example of mixed structured/unstructured meshes created automatically within OMNIS.

What project are you most proud of and why?

I am proud of all the innovations NUMECA has been able to develop and bring forward to the market, and in particular these firsts, among others, in reference to breakthrough we introduced as world premiere on the market:

  • Full Automatic Block structured grid generation for Turbomachinery with AutoGrid, that is today the world reference.
  • Automatic Unstructured Full Hex grid generation with Hexpress. As far as I know, we are the only one having developed such tool, combing the power of full Hex with unstructured grids.
  • Fastest solvers on the market with remarkable speed-up over competition on similar hardware, with 5x to 15x speed-up in turbomachinery and external aerodynamics.
  • A unique technology based on solving the equation in the frequency domain for unsteady rotating machinery, called NLH, offering a speed-up of 100x to 1000x compared to other methods and having no equivalent as off today.
  • A multidisciplinary client-server new software environment encapsulating all CFD codes and tools in One single unique user experience: OMNIS. As far as I know OMNIS has no equivalent on the market, in terms of integration and versatility.
Examples of unstructured mesh created in OMNIS – Full car external aero, under hood and engine bay.

Are you reading any interesting books we should know about?

After several visits to ancient sites in Greece, I began reading Aegean Art by Jean-Claude Poursat. It is amazing to see how the Minoan and Mycean ancient civilizations used art, culture and technology during the age of pottery and bronze, in a yet modern, creative and well-organized civilization.

What software or tools do you use every day?

E-Mail.

[John] Don’t we all. It would be so much easier to be in the office and just walk down the hall to ask a question. Soon, hopefully soon.

What do you do outside the world of CFD?

Fluid is present all around us. And as we live in a fluid, I cannot really escape from CFD. But I love many other things such as sailing, travelling, and reading.

What is some of the best CFD advice you’ve ever received?

  • Make high quality grids! Otherwise you will pay the price of poor accuracy and bad CPU performance.
  • Focus on the physics you want to capture, neglect secondary effects that have no impact on the solution and that could largely complexify the entire workflow.
  • Do validation, comparing to reference test cases or analytical solutions.

[John] Glad to see you listed grids first. The right mesh, a suitable mesh for what you’re trying to accomplish, is key. And it’s also where our users have the most direct impact on their results.

If you had to pick a place to have dinner, where would you go?

Brussels is full of nice restaurants but here a some of my favorites.

[John] Marc, I hope we have a chance to share a meal at any of these places soon. But reading the menus was a mistake because now I’m starving. Reading the dessert menus was an even bigger mistake because of my sweet tooth.

This entry was posted in People & Places, Software and tagged , , , , , , . Bookmark the permalink.

1 Response to I’m Marc Tombroff and This Is How I Mesh

  1. Pingback: This Week in CFD - Another Fine MeshAnother Fine Mesh

Leave a Reply